the allosteric three-site model for the ribosomal elongation cycle

THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 26, Issue of September 15, pp. 13103-13111,1988 Printed in U.S.A.

The Allosteric Three-site Model for the Ribosomal Elongation Cycle NEW INSIGHTS INTO THE INHIBITION MECHANISMS OF AMINOGLYCOSIDES, THIOSTREPTON, AND VIOMYCIN*

(Received for publication, January 25, 1988)

Thomas-Peter Hausner, Ute Geigenmiiller, and Knud H. Nierhaus From the Man-Planck-Znstitut fur Molekulare Genetik, Abteilung Wittmann, Zhnestrasse 73, 0-1000 Berlin-Dahlem, West Germany

According to the allosteric three-site model for the ribosomal elongation cycle (Rheinberger, H. J. and Nierhaus, K. H. (1986) J. Biol. Chem. 261, 9133- 9139), two types of A site (aminoacyl-tRNA site) occupation exist. First is the A site occupation after initiation (i-type), with only one site, the P site (peptidyl- tRNA site), being prefilled with a tRNA (initiator tRNA). Second is the A site occupation after an elongation cycle (e-type), with two prefilled sites, namely the P and E sites containing peptidyl-tRNA and deacylated tRNA, respectively. The individual reactions of the elongation cycle were tested, including both types of A site occupation in the presence of various antibiotics. A test system was used allowing the functional studies to be made with quantitative tRNA binding at 6 mM M&+.

The following results were obtained: 1) thiostrepton (5 X lo-‘ M) induced a complete block of both EF- (elongation factor) G dependent and EF-G independent translocation, in agreement with older observations. The A-site occupation of the e-type was severely inhibited in contrast to that of the i-type. Thus, thiostrepton blocks the allosteric transitions in both directions, i.e. the transition from pre- to post-translocational state (translocation) and that from the post- to the pretranslocational state (A site occupation of the e-type). In addition the ribosomal binding of EF-G*[’H] GMPPNP was inhibited by about 60%. 2) Similarly, viomycin (5 X lo-’ M) appears to be an inhibitor of both allosteric transitions, since it strongly inhibited the e-type (but not the i-type) A site occupation in addition to translocation. 3) The aminoglycosides streptomycin, hygromycin B, neomycin, kanamycin, and gentamicin prevented A site occupation of the e- type (residual activity below 15%). Neomycin and hygromycin, in addition, blocked the translocation reaction. Only marginal effects were observed with A site occupation of the i-type. It appears that the inhibition of the A site binding of the e-type (allosteric transition from the post- to the pretranslocational state) is the predominant effect of the misreading-inducing aminoglycosides.

Analysis of the tRNA-binding capacities of ribosomes and the functional steps of the elongation cycle has led to an allosteric three-site model (1, and references herein). This allosteric three-site model for the ribosomal elongation cycle

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

n E P A

Tronsiacotloo

1

FIG. 1. The two types of A site occupation according to the allosteric three-site model. The i-type occupation follows the initiation reactions, the E site remaining unoccupied. The e-type occupation occurs after an elongation cycle, and is characterized by an occupied E site.

is characterized by three tRNA-binding sites A,’ P, and E. Two of these sites (A and E) are allosterically linked via a negative cooperativity, i.e. occupation of one site decreases the tRNA affinity of the other and vice uersa. Thus, the model defines two states of the elongating ribosome, the pre-translocational state, with A and P as the high-affinity sites, and the post-translocational state, with P and E as high-affinity sites. Both respective high-affinity sites are occupied by tRNAs before and after translocation, and both tRNA9 simultaneously undergo codon-anticodon interaction.

The model implies that two types of A site occupation have to be distinguished (see Fig. 1). The first is the A site occupation of a complex equivalent to the 70 S-initiation complex with only the P site prefilled with a tRNA. This type of occupation we define as i-type (i for initiation). On the other hand an A site occupation following an elongation cycle involves a ribosomal complex carrying two tRNAs, uiz. a peptidyl-tRNA and a deacylated tRNA at the P and E sites, respectively. This type of occupation we define as e-type (e for elongation). i- and e-types of A site binding are function- ally not equivalent, since the i-type of A site occupation occurs

The abbreviations used are: A site, aminoacyl-tRNA site: P site, peptidyl-tRNA site; E site, exit site specific for deacylated tRNA; AcPhe-tRNAPhe, N-acetyl-Phe-tRNAPhe; EF, elongation factor; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; GMPP- NP, guanyl-5’-yl-imidodiphosphate.

13103

13104 Inhibition Mechanisms of Aminoglycosides, Thiostrepton, and Viomycin

effectively at 0 "C whereas the e-type does not (1). In this paper we demonstrate that the allosteric three-site

model can improve and deepen our understanding of the inhibition mechanism of some antibiotics. In particular we show that viomycin blocks the allosteric transitions between the pre- and the post-translocational states in both directions rather than being a translocation inhibitor per se (2, 3). Thiostrepton displays a similar inhibition pattern. Further- more, the aminoglycosides inducing misreading, which have been found to affect a variety of ribosomal functions (reviews, 4-7), preferentially block one reaction of the elongation cycle, namely A site binding of the e-type.

MATERIALS AND METHODS

The isolation of tightly coupled 70 S ribosomes from Escherichia coli K12, strain D10, was described in Ref. 1. 1 A~M) unit of 70 S

tRNA to be equivalent to 1500 pmol. Poly(U) and tRNAPh" were ribosomes was taken to be equivalent to 24 pmol, and 1 A260 unit of

purchased from Boehringer Mannheim. AcPhe-tRNA and Phe-tRNA were prepared according to Ref. 8 and were freed of deacylated tRNA by chromatography on benzyolated DEAE-cellulose (9) with the modification of Ref. 10. Isolation of EF-Tu and EF-G followed the procedure of Ref. 11. Samples containing both 3H- and "C-isotopes on wet nitrocellulose filters were treated with 0.3 ml of HzO and 8 ml of Filter-Count (Packard), and the samples were counted after complete dissolution of the filters in an LKB 1219 Rackbeta, using the program "Dual Label." Separation of the counts originating from the respective isotopes was better than 99.7%.

The antibiotics were obtained from Boehringer Mannheim (chloramphenicol, kanamycin-sulfate (kanamycin A), neomycin-sulfate (neomycin B), and tetracycline-HCl), Calbiochem (hygromycin B), Heyden, Munchen (thiostrepton), Pfizer, Tokyo (viomycin), Serva (spectinomycin-HC1, streptomycin-sulfate), and Sigma (gentamicin- sulfate (gentamicin C), lincomycin-HC1). a-Sarcin was a kind gift from I. Wool, University of Chicago and J. Davies, Institut Pasteur, Paris. Stock solutions of the drugs contained 10 mM Tris, the pH being adjusted to 7.8. The tetracycline solution was adjusted to pH 7.8 and used immediately. Exceptions were chloramphenicol which was dissolved in 50% ethanol, and thiostrepton, which was dissolved in 10% (v/v) MezSO. All stock solutions contained the respective drugs at a concentration 40 times higher than the final concentrations given in Table I. When thiostrepton was present the assay contained 0.25% (v/v) MezSO, and in the case of chloramphenicol 1.25% ethanol. Accordingly, extra controls (minus thiostrepton and minus chloramphenicol) were performed with assay mixtures containing the same concentrations of MezSO and ethanol, respectively; the results of these controls were identical to that of the standard control (minus

The test system for tRNA binding, translocation, and the puromycin reaction followed the procedure of Ref. 9, with the exception that the ionic concentrations were 20 mM Hepes-KOH, pH 7.8,6 mM magnesium acetate, 150 mM NH4Cl, 0.6 mM spermine, and 0.4 mM spermidine in all steps. Further details are given below. Where not otherwise indicated, the deviation of the individual values from the average value was less than 5%.

Poly(U)-dependent Poly(Phe) Synthesis-The inhibitory effects of the various antibiotics listed in Table I were studied over a range of concentrations between lo" and 5 X lo-* M. Binding of Ac13H]Phe- tRNA (5024 dpm/pmol) to poly(U) programmed 70 S ribosomes was performed as described (12) under the ionic conditions given above (molar ratio AcPhe-tRNAl70 S = 2:l). The charging mixture, containing the same ionic concentrations together with ["CIPhe, was added and, after addition of the antibiotics, poly(Phe) synthesis was allowed to proceed for 10 min at 37 "C. The trichloroacetic acid precipitate was collected on glass filters and counted. Each aliquot contained 5.1 pmol of 70 S ribosomes. In the control (minus drug) 16 Phe molecules (["CC]Phe-tRNA = 11.9 dpm/pmol) were incorporated statistically per ribosome, and 26% carried an A c [ ~ H ] P ~ ~ residue. Thus, the incorporation per active ribosome was 63 Phe residues.

P Site Binding and Puromycin Reaction (Table I)-Each aliquot contained 11.1 pmol of poly(U)-programmed 70 S ribosomes and the indicated amounts of antibiotics. The molar ratio of added Ac[I4C] Phe-tRNA/70 S was 0.761, at a specific activity of 980 cpm/pmol. After the binding incubation (30 min at 37 "C, followed by removal of two aliquots), a translocation incubation (10 min at 37 "C) was

drug).

carried out in the presence and absence of EF-G. When present, EF- G was added in nearly stoichiometric amounts (EF-G/7O S = 0.81). The puromycin reaction was performed at 0 'C for 90 min. In the control (minus drug) 0.57 AcPhe-tRNA molecules were bound per 70 S ribosome, 94% of which was present in the P site, according to the puromycin reaction. The values were calculated as follows. The relative P site binding is given by the expression B'p/Bp X 100, where Bp denotes binding to the P site, and B'p, the corresponding binding in the presence of an inhibitor. The relative peptidyltransferase activity is given by [(PM'/B'p)/(PM/Bp)] X 100, where PM is the puromycin-sensitive fraction of the AcPhe-tRNA bound to the P site in the absence of EF-G, and PM' is the corresponding value in the presence of an inhibitor.

A Site Binding and EF-G-dependent Translocation (Fig. 2)"Each aliquot contained 10.3 pmol of poly(U)-programmed 70 S ribosomes. The P sites, or the P and E sites, were filled with deacylated tRNAPhe (added tRNA/70 S = 1.5:l and 5:1, respectively). After addition of the antibiotics A c [ ~ H ] P ~ ~ - ~ R N A ' ~ ' was added (AcPhe-tRNAl70 S = 0.4:l; specific activity 3401 cpm/pmol). The sample was treated as described for P site binding in the preceding section.

The relative A site binding was calculated according to B'a/Ba X 100, where Ba represents binding to the A site, and B'a is the binding in the presence of an inhibitor.

The relative translocation reaction was corrected for the inhibition of both the A site binding and the puromycin reaction according to the following formula:

where APMa = PMa+c - PMa-c, PMa-c being the puromycin- sensitive fraction of the AcPhe-tRNA bound to the A site and then subjected to a translocation incubation in the absence of EF-G, and PMa+c being the puromycin-sensitive fraction with a translocation incubation in the presence of EF-G. APM'a is the corresponding value in the presence of an inhibitor. For more explanations see legend to Table 3 in Ref. 13.

EF-G-independent (Spontaneous) and EF-G-dependent Transloca- tion of AcPhez-tRNA (Fig. 3)"Each aliquot contained 15 pmol of poly(U)-programmed ribosomes. A c [ ' ~ C ] P ~ ~ - ~ R N A was bound to the P site (AcPhe-tRNA/70 S = 0.5:l; specific activity 555 dpm/pmol) by incubating for 10 min at 37 "C. The remaining nonfilled P sites were occupied with deacylated-tRNA (tRNA-Phe/70 S = 0.8; 5 min, 37 "C). 0.37 AcPhe-tRNA were bound/70 S ribosome; 100% was found at the P site according to the puromycin reaction. Next, ['HIPhe- tRNA was added (Phe-tRNAI70 S = 0.25:l; specific activity 4707 dpm/pmol). After an incubation for 5 min at 0 "C (0.15 Phe-tRNA were bound/ribosome), the antibiotics were added as indicated, and a second incubation was made at 37 "C for 10 min in the presence and absence of EF-G (EF-G/70 S = 0.7). The puromycin reaction was performed for 90 min at 0 "C. The relative activities observed in the puromycin and translocation reactions were calculated as described above.

Binding of Deaeylated r4C]tRNA to the E Site (Table ZZ)-Each aliquot contained 20 pmol of poly(U)-programmed tightly coupled ribosomes. The P sites were closed with nonlabeled tRNAPhe (tRNA/ 70 S = 1.5:1), and then the antibiotics were added followed by an incubation for 5 min at 37 'C. After addition of deacylated ["C] tRNAPhe (tRNAPhe/70 S = 3.4:1, specific activity 145 dpm/pmol), the sample was incubated for 10 min at 37 "C and the binding determined. The control (minus drug) revealed a binding of 0.84 ["C]tRNAPhe/ ribosome. The deviation of the individual values from the average value was less than 5%.

Binding of EF-G. PHJGMPPNP to Ribosomes in the Absence and Presence of Thiostrepton (Fig. 5)"Each aliquot contained 12 pmol of poly(U)-programmed ribosomes. In the first incubation the P sites were closed with deacylated tRNAPhe (tRNA/70 S = 1.2:l); in the second step the E sites were filled (tRNA/70 S = 3.5:l). In experiment 1 of Fig. 5, EF-G (EF-G/70 S = lOl), [3H]GMPPNP (GMPPNPI70 S = 272, specific activity 2424 dpm/pmol), and (when indicated) thiostrepton were added after the first step. The sample was incubated (2 min at 37 "C) and the binding determined on nitrocellulose filters. In experiment 2 EF-G, [3H]GMPPNP, and (as indicated) thiostrepton were added after the second step and the sample processed as above. In experiment 3 Ac["C]Phe-tRNA was added after the second step (AcPhe-tRNA/70 S = 0.94:1, specific activity 564 dpm/pmol) and the sample incubated for 30 min at 37 "C. A binding control

Inhibition Mechanisms of Aminoglycosides, Thiostrepton, and Viomycin 13105

TABLE I Drug effects on poly(Phe) synthesis, P site binding, and the

puromycin reaction Poly(Phe) synthesis: control (minus antibiotic) 100% = 994 dpm

(16 Phe/ribosome), background 51 dpm. P site binding: control (minus drug) 100% = 6,231 cpm, background 255 cpm. Puromycin reaction: control (minus drug) 100% = 2,619 cpm, background 157 cpm; the values were corrected for P site inhibition (see "Materials and Methods").

Activity Antibiotics Poly(Phe) P site Puromycin

synthesis binding reaction

[Io-K M] % Control (minus drug) 100 100 100 Control drugs

Tetracycline 37.5 47 98 86 Chloramphenicol 50 18 100 5 a-Sarcin 4 10 100 100

Lincomycin 50 63 91 8 Thiostrepton 0.5 1 95 100 Viomycin 5 3 100 70 Aminoglycosides

Streptomycin 50 15 100 71 Hygromycin B 5 3 100 81 Neomycin B 5 1 100 79 Kanamycin A 5 2 100 96 Gentamicin C 5 7 100 90

revealed that 0.3 AcPhe-tRNA were bound/70 S. Then EF-G, [3H] GMPPNP, and (where indicated) thiostrepton were added, and the sample was processed as above. Background values (minus ribosomes, 621 dpm) were subtracted. The data are averages of triplicate deter- minations, and the standard deviation is given.

Kinetics of the Binding of the Ternary Complex r3H1Phe-tRNA. EF-Tu.GTP to the A Site (Fig. &-Each aliquot contained 11.1 pmol of poly(U)-programmed 70 S ribosomes. For A site binding of the i- type (Fig. 4A), Ac["C]Phe-tRNA was bound to the P site (AcPhe- tRNA/70 S = 1:1, specific activity 588 dpm/pmol). A binding control indicated a binding of 0.66 AcPhe-tRNA/70 S, and a puromycin control showed that 94% of the bound material was present at the P site. After addition of the drugs and an incubation at 37 "C for 2 min, the ternary complex [3H]Phe-tRNA.EF-Tu. (GTP or GMPPNP) was added (ternary complex/7O S = 0.31:l; specific activity 4949 dpm/ pmol), and the mixture incubated at 37 "C. Aliquots were withdrawn after the times indicated and the binding assessed. Backgrounds (minus ribosomes, 438 dpm) were subtracted. The points given in Fig. 4 represent the bound 3H material.

For A site binding of the e-type (Fig. 4B), the P site was closed with deacylated tRNAPhe (tRNA/70 S = 2:1), and the Ac[14C]Phe- tRNA was bound to the A site (AcPhe-tRNA/70 S = 1:1, specific activity 584 dpm/pmol; 30 min at 37 "C). A binding control indicated that 0.4 AcPhe-tRNA bound/7O S, and the puromycin control that 89% was located at the A site. A translocation reaction was then carried out (EF-G/70 S = 0.8:1), and a further puromycin control showed that 100% of the AcPhe-tRNA was now at the P site. Where indicated, the mixtures received the antibiotics, after which the ternary complex (specific activity of [3H]Phe-tRNA: 10,830 dpm/ pmol) was added and the sample processed as described above. The background minus ribosomes (2284 dpm) was subtracted.

The 100% values for the binding of [3H]Phe residues after incubation of 1, 2, and 5 min were as follows: for i-type binding (Fig. 4A) 0.14,0.18, and 0.18 bound Phe residuesl70 S in the presence of GTP, and 0.14, 0.16, and 0.17 Phe residues, respectively, in the presence of GMPPNP. For e-type binding (Fig. 4B), the corresponding values were 0.2, 0.25, and 0.27 in the presence of GTP, and 0.05, 0.10, and 0.14, respectively, in the presence of GMPPNP.

RESULTS

Outline of the Experimental Procedures The antibiotics used in this study were first tested for their

ability to inhibit poly(Phe) synthesis over a concentration range of to 5 x M. Wherever possible concentrations for the further assays were chosen such that the inhibition of

poly(Phe) synthesis was at least 50% (Table I). Only lincomycin showed a weak effect on the poly(Phe) synthesis (63% residual activity at 5 X M). However, the well-documented blocking induced by lincomycin of the puromycin reaction with AcPhe-tRNA (review, 4, 5) could be reproduced (Table I, 8% residual activity).

The main reactions of the elongation cycle (binding to A and P sites, respectively, peptidyl transferase reaction and translocation) were studied using AcPhe-tRNA (Table I, Fig. 2). This analogue of a peptidyl-tRNA binds directly to the P sites of poly(U)-programmed ribosomes and only to the A sites if the P sites were filled with deacylated tRNAPhs before. This system is well suited for the analysis of the reactions of the elongation cycle (14). In addition translocation and A site occupation were analyzed under more physiological conditions. In the case of translocation, AcPhe2-tRNA was formed at the A site, and its translocation was assessed in the presence of EF-G. In the absence of EF-G, the spontaneous translocation was tested (Fig. 3; for a detailed analysis see Ref. 14).

For the analysis of the A site binding, the P sites were preoccupied with AcPhe-tRNA before the kinetics of the binding of ternary complexes were monitored (Fig. 4). The e- type binding of the ternary complex was assayed as follows: P and A sites were occupied with deacylated tRNA and AcPhe-tRNA, respectively. Then the tRNAs were translo- cated to E and P sites, respectively, before adding substoichiometric amounts of ternary complexes. Under these conditions AcPhe2-tRNA is formed nearly exclusively: indicating that the ternary complexes bind with high priority to those ribosomes which carry an AcPhe-tRNA at the P site. The binding of the ternary complex was performed either with GTP (Fig. 4, circles) or GMPPNP (triangles). In the presence of GTP, the binding is followed by the formation of AcPheP- tRNA, shifting the reaction toward A site occupation and thus possibly masking an impairment of the A site binding in our test system. In fact whenever we see a reduced A site binding the inhibition is more pronounced in the presence of GMPPNP. It follows that the target reaction(s) can be iden- tified with the experiments of Table I and Fig. 2, whereas those of Fig. 4 demonstrate overall effects on the elongation cycle within the time resolution of the kinetics.

In addition the E-site occupation was analyzed (Table 11). In the course of saturating the three ribosomal-binding sites with deacylated tRNA, the second deacylated tRNA binds to the E site (1,15,16). Accordingly, the P site was blocked with nonlabeled tRNA, then the binding of deacylated ["C] tRNAPhe to the E site was measured. Table I1 demonstrates that all the antibiotics (including those which effectively prevent the A site binding of the e-type, see Fig. 2) hardly impeded the E site binding, with the surprising exception of the A site-specific tetracycline.

Testing the Specificity of the Fractional Tests by Some Control Drugs-A number of "control drugs'' with known and defined inhibition mechanisms were included to test the specificity of the measurements. Tetracycline, which specifically inhibits A site binding (17) as long as substoichiometric amounts of A site ligands are applied (18), affects A site occupation of both the i- and e-type (Fig. 2; 24 and 35% residual activity, respectively, tested with substoichiometric amounts of AcPhe-tRNA; and Fig. 4). Chloramphenicol, the classical inhibitor of peptidyl transfer (19, 20), blocked the puromycin reaction (Table I, 5% residual activity) preventing a further analysis of the translocation reaction. The A site occupation of the e-type was affected weakly (78% activity, Fig. 2B). a-Sarcin, which cleaves one phosphodiester bond in

* H.-J. Rheinberger and K. H. Nierhaus, unpublished observations.


A

B

A-site binding of the i-type and translocation

- drug r m’nus control drugs ommcqlyroslder

A-site binding of the e-type and trans\ocation

%

c o n t r d drup - 12 11 11 ~ q q

- E l m LIN THI VI0 St4 HYG NE0 KAN GEN

~

n.d I nd. n d nd. n.d n.d n d

FIG. 2. A site binding of AcPhe-tRNAPb‘ and the subsequent translocation reaction at 37 “C. A, A site binding of the i-type: control (minus drug) 10,599 cpm = loo%, the background of 296 cpm being subtracted from the values. Translocation: control (minus drug) 6,447 cpm = loo%, background 650 cpm; the values were corrected for inhibition of A site binding and peptidyltransferase activity (see “Materials and Methods”). B, A site binding of the e-type: control (minus drug) 4,889 cpm = loo%, the background of 362 cpm being subtracted from each value. Translocation: control (minus drug) 3,960 cpm = loo%, background 138 cpm. The values were calculated as described under “Materials and Methods.” CON, control; TET, tetracycline (37.5 X lo-’ M); SAR, a-sarcin (4 X lo-’ M); LIN, lincomycin (50 X lo-’ M); THI, thiostrepton (0.5 X lo-‘ M); VIO, viomycin (5 X lo-’ M); S M , streptomycin (50 X M); HYG, hygromycin B (5 X lo-‘ M); NEO, neomycin B (5 X M); KAN, kanamycin A (5 X low6 M); GEN, gentamicin C (5 X M). e, ribosome carrying a deacylated tRNA at the

U p site; 8, pretranslocational ribosome carrying a deacylated tRNA at the P site and a N-acetyl-Phe-tRNA at

the A site; 8, post-translocational ribosome carrying a N-acetyl-Phe-tRNA at the P site and a deacylated tRNA

at the E site; y , deacylated tRNA B, N-acetyl-Phe-tRNA; I , Phe-tRNA.

+ t FIG. 3. EF-G-dependent and EF-

G-independent (spontaneous) translocation of AcPhel-tRNA. 100% values correspond to 7150 and 1479 dpm for EF-G-dependent and independent translocation, respectively. Background values (minus puromycin) have been subtracted, and corrections were made as described under “Materials and Meth- ods.’’ Abbreviations and symbols are explained in the legend to Fig. 2.

minus

,“

YJ u. x., !i

CON THI VI0 SH HYG NED KAN GEN

yB - EF-G

23 S rRNA (21, 22) was included. The cleavage site in E. coli are inhibited, whereas all other intrinsic ribosomal functions (after residue GZBB1) is probably located at the overlapping remain unaffected (see Ref. 22 for details and discussion). Fig. binding sites of the elongation factors, since the binding of 2 demonstrates that EF-G-dependent translocation is severely both factors and, consequently, the factor-related functions inhibited (14% residual activity), while the other functions

Inhibition Mechanisms of Aminoglycosides, Thiostrepton, and Viomycin 13107 TABLE I1

Drug effects on the E site occupation 100% activity corresponded to 2435 dpm.

Activity

M] % Control (minus drug) 100 Control drugs

Tetracycline 37.5 60 Chloramphenicol 50 84 a-Sarcin 4 100

Lincomycin 50 89 Thiostrepton 0.5 75 Viomycin 5 76 Aminoglycosides

Streptomycin 50 82 Hygromycin B 5 14 Neomycin B 5 92 Kanamycin A 5 74 Gentamicin C 5 89

tested were not affected. Thus, under the conditions used the defined inhibition pattern of tetracycline, chloramphenicol, and 0-sarcin could be reproduced.

Inhibition Patterns of Various Antibiotics

Lincomycin-Lincomycin, the well-documented inhibitor of the peptidyl transferase reaction (review 4, 5), blocks the puromycin reaction completely under our conditions (Table I, 8% activity). This prevented a further analysis of the translocation reaction. In addition lincomycin reduced A site binding of the e-type using AcPhe-tRNA as the A site ligand (51% residual activity, Fig. 2B) but showed only a marginal effect on A site binding of the i-type (83% residual activity,

Fig. 2A). However, under more physiological conditions (binding of ternary complex) the A site binding of both types was hardly affected (Fig. 4, A and B) .

Thiostrepton-Thiostrepton completely abolished the EF- G-dependent translocation activity (Fig. 2, A and B ) and was the strongest inhibitor of this function among all the drugs tested. Interestingly, the spontaneous EF-G-independent translocation was also completely inhibited (Fig. 3). This indicates that the ribosomal capability for translocation per se is affected and not merely an EF-G-mediated process. However, thiostrepton reduced the binding of EF-G. [3H] GMPPNP to various functional states by about 60%, i.e. the inhibition was independent of the number of tRNAs prebound to the ribosomes (Fig. 5, about 40% residual activity). Fur- thermore, A site occupation of the e-type is strongly impaired (32% activity, Fig. 2B) and that of the i-type is also somewhat weaker (69%, Fig. 2A). Under more physiological conditions the same pattern was found; uiz. a nearly complete block of the e-type binding (Fig. 4B), a weaker effect on the i-type binding (Fig. 4A). P site binding and the puromycin reaction are in contrast not affected (Table I).

Viomycin-Viomycin effectively prevents translocation (Figs. 2 and 3, 0-22% activity), in agreement with previous observations (2, 3). Surprisingly, it impedes the e-type A site binding to a comparable degree (12% activity, Fig. 2B) , whereas the i-type A site binding and the puromycin reaction are only moderately reduced (67 and 70% activity, Fig. 2A and Table I, respectively). It follows that a ribosome in a post-translocational state (P and E sites filled with tRNA) cannot be transferred back to the pretranslocational state (via A site binding of the e-type) in the presence of viomycin. Similarly, a ribosome in the pretranslocational state (A and P sites occupied with tRNA) cannot move to the post-translocational state (translocation) in the presence of the drug.

All the aminoglycosides tested here dramatically reduced A site binding of the e-type (6-11% activity, Fig. 2B). In contrast

P A

1u.y

20

1 2 5 1 2 5 1 2 5 1 2 5 1 2 5 h m e l m n l

FIG. 4. Kinetics of the binding of the ternary complex to the A site. M, the ternary complex [3H] Phe-tRNA-EF-Tu-GTP containing GTP; A-A, the ternary complex with GMPPNP. A, A site binding of the i-type. B, A site binding of the e-type. Abbreviations and symbols are explained in the legend to Fig. 2. For further details see "Materials and Methods."


I / ”

P P

- 776 t 290 100

IF-G. P H I G H P R I P + l3L f 10 Ll

I W G M P P N P

FIG. 5. The binding of EF-G*[3H]GMPPNP to ribosomes in the presence or absence of thiostrepton (THI). The symbols are explained in the legend to Fig. 2. For details see “Materials and Methods.”

A site binding of the i-type (73-92%, Fig. 2A; see also Fig. 4A) , and the puromycin reaction (71-96% activity, Table I) were hardly affected. The reduction of the e-type A site binding was so pronounced that the subsequent translocation reaction could not be determined (n.d. in Fig. 2B) . However, the translocation reaction following the i-type A site binding could be assessed. Fig. 2A demonstrates that neomycin and hygromycin B strongly impair the translocation reaction (9 and 33% activity, respectively), the other drugs showing weaker effects (66-89% activity). However, even these weaker effects are significant, as revealed by more sensitive translocation assays, uiz. EF-G-dependent translocation of AcPhe- tRNA at 0 “C (data not shown) or spontaneous EF-G-independent translocation of AcPhe2-tRNA at 37 “C (Fig. 3). In this assay system all the aminoglycosides tested induced a severe block of translocation. It follows that the post- to pretranslocational transition (A site binding of the e-type) is heavily impaired by these aminoglycosides, whereas variable effects are found for the reverse transition (translocation).

DISCUSSION

The Allosteric Three-site Model and the System Used

Table I11 summarizes the features of the allosteric three- site model, based on a number of observations made during the last six years (1,15,16,23,24). Accordingly, the elongation cycle has been sketched with “rectangular” and “round” ribosomes, symbolizing the pre- and the post-translocational states (Fig. 1, see also Ref. 1). Fig. 1 also illustrates that a ribosome entering the elongation cycle after the initiation reaction carries only one tRNA at the P site, whereas at the end of an elongation cycle the ribosome carries two tRNAs at the P and E sites. We term the corresponding two types of A site binding i-type and e-type, respectively, as already mentioned.

In all the experiments described here, a system was used which contains 6 mM Mg”’, 150 mM NH,+, 0.6 mM spermine, and 0.4 mM spermidine. We have shown that under these conditions (6 mM M$+) the ribosome quantitatively binds tRNA (25), whereas at 3 mM M$+, but under otherwise identical conditions, the rate and accuracy of poly(Phe) synthesis approach the corresponding in uiuo values ( 1 2 ) , and the tRNA binding is still nearly quantitative. At both 3 and 6 mM Mg2+ the E site is functional, and the features compiled in Table I11 are valid (25).

Recently, we used a heteropolymer mRNA to check the validity of the allosteric three-site model, where the binding

TABLE 111 Features of the allosteric three-site model

1. A ribosome contains three tRNA-binding sites A, P, and E. In the course of elongation a tRNA occupies the sites successively in the order A -+ P + E. The E site exclusively binds deacylated tRNA.

2. A and E sites are allosterically linked via negative cooperativity, i.e. occupation of one site decreases the affinity of the other and vice versa.

Consequences: An elongating ribosome can adopt two states: -The pre-translocational state, where A and P

-The post-translocational state, where P and E

Deacylated tRNA is released upon A site occupation and not during translocation.

sites have high affinities for tRNA.

sites have high affinities.

3. Both tRNAs present on the ribosome before and after translocation undergo codon-anticodon interaction simultaneously.

4. Both elongation factors EF-Tu and EF-G promote the transition from one state to the other: EF-Tu that of the post- to the pretranslocational, and EF-G that of the pre- to the post-translocational transition.

of tRNAs could be assigned unequivocally to the various sites depending on the respective codons. The allosteric three-site model could be fully ~onfirmed.~

In a recent report the allosteric interplay between A and E sites was questioned, and the authors concluded that EF-G- dependent GTPase is necessary for efficient tRNA release (i.e. the A site occupation of the e-type has no significance) and that the E site is expected to have a low affinity during elongation in vivo due to the low M e concentrations in the cell (26). The latter conclusion has already been shown to be a misinterpretation, since under the conditions used by the authors all three sites A, P, and E are inactivated in a coordinated fashion (25). Furthermore, numerous EF-G. GTP-dependent translocation reactions have been published where the translocation is not accompanied by a tRNA release, including conditions with 6 and 3 mM Mg”’ (17,23-25). Clearly, EF-G-dependent GTPase is neither required for nor involved in tRNA release during elongation.

Inhibition Patterns of the Antibiotics Tested (Not Including the Aminoglycosides)

The drug effects compiled in Table I and Fig. 2 can be compared directly, since they were collected under identical or equivalent conditions (with the exception of the poly(Phe) synthesis, Table I). In contrast the spontaneous EF-G-independent translocations at 37 “C (Fig. 3) are more sensitive to the antibiotics and can therefore only be compared qualita- tively with the experiments of Fig. 2.

The control drugs tetracycline, chloramphenicol, and a- sarcin show their expected inhibition pattern (Table I, Figs. 2 and 4, see “Results”), and thus indicate the specificity of the test systems. An interesting detail was observed with tetracycline. Among the numerous drugs which effectively prevented the A site binding of the e-type (Fig. 2B) , the more or less A site-specific tetracycline was the only one which significantly reduced the binding to the E site (Table 11). However, the inhibition of tRNA binding to the E site is much less pronounced than that to the A site (40% versus 76% inhibition, compare Table I1 with Fig. 2A), and the inhibition of the A site binding exceeds 70% even at high

A. Gnirke, U. Geigenmuller, H.-J. Rheinberger, and K. H. Nier- haus, unpublished observations.

Inhibition Mechanisms of Aminoglycosides, Thiostrepton, and Viomycin 13109

AcPhe-tRNA concentrations (about 3 molar excess over ribosomes (18)). The interpretation is that tetracycline at least partially mimics a tRNA at the A site, thus establishing a pretranslocational state. This state is characterized by a low E site affinity (Table I11 and Fig. 1 in Ref. 1). Recently, this interpretation could be verified by using a heteropolymer mRNA and the corresponding codon-specific ~ R N A s . ~

Lincomycin-Lincomycin is a well-known inhibitor of the peptidyltransferase activity in model reactions ((27), 8% activity in Table I). In addition it significantly inhibits A site binding of the e-type (51% activity) in contrast to that of the i-type (83%, Fig. 2). The apparently low inhibition of the poly(Phe) synthesis (63%) could be due to a chasing of the drug by Phe-tRNA, an effect which is more or less specific for this acyl-tRNA. (Possibly, the same explanation holds true for the marginal effects of chloramphenicol on the poly(Phe) synthesis at concentrations below 5 X M). Interestingly, lincomycin diminishes the affinity labeling of protein L2 by bromoacetyl-Phe-tRNA (28). L2 is one of the most likely candidates for the peptidyltransferase (for discussion see Ref. 29). The fact that the synthesis of peptidylpu- romycin is not impeded by lincomycin (30) is due to the nascent peptides which prevent the binding of the drug (31).

Spontaneous as well as EF-G dependent translocation is completely blocked by thiostrepton (Figs. 3 and 2), in good agreement with previous observations (32). In addition a severe block of the A site binding of the e-type (Fig. 4B) is observed, which predominates over the corresponding inhibition of the i-type A site binding (Fig. 4A) and also over the inhibition of the EF-G binding (Fig. 5 ) . We are inclined to interpret the block of either allosteric transition as the primary inhibition mechanism of the drug. In this view thio-

U. Geigenmuller and K. H. Nierhaus, unpublished observations.

FIG. 6. The three functional phases of the ribosome. Initiation, the elongation cycle comprising the reactions (El-5) and the termination. The scheme is an improved version of that presented in Ref. 6, the symbols for the ribosome in the pre- and post-translocational state being taken from Ref. 1. Actions of those antibiotics not considered here are found in Refs. 4-6.

strepton freezes either the pre or the post-translocational state, thus blocking the following allosteric transition. AC- cording to the three-site model an allosteric transition is a prerequisite for a factor-dependent GTPase activity. There- fore, the reported inhibition of factor-dependent GTPase activity appears to be rather a consequence than a cause of the inhibitory effects of thiostrepton (for this and other effects of the drug see Ref. 5 ) .

Viomycin-Viomycin is a well-characterized inhibitor of translocation (2, 3, 13). In addition viomycin disturbs the accuracy of tRNA selection more heavily than streptomycin (13). Our analyses showed that both translocation (15% activity, Fig. 2A) and A site binding of the e-type (12%, Fig. 2B) are equally strongly reduced. The other functions tested are affected much less, if at all (70 and 67% for the peptidyltransferase activity and A site binding of the i-type, respectively, see Table I and Fig. 2A). It thus appears that viomycin is an inhibitor of both allosteric transitions and freezes either state of the elongating ribosome, i.e. the pre-translocational as well as the post-translocational state. Therefore, viomycin acts similarly to thiostrepton with regard to the primary inhibition mechanism but differs in its secondary effects (e.g. misreading).

The Inhibition Patterns of Aminoglycosides The aminoglycosides are the classical inducers of misread-

ing effects. The misreading aminoglycosides are almost unique in their bactericidal activity, since most of the protein-synthesis inhibitors are bacteriostatic. These aminoglycosides inhibit poly(Phe) synthesis, in contrast to those which do not misread (such as spectinomycin and kasugamycin), which have structural peculiarities and are not bactericidal. The latter should therefore not be included in the aminoglycoside

VIOAYCIN 1 N E O N Y C l d

AHINOGLYCOSIDES STREPIONYCIN H Y G R O I W I N B

KANAPIYC I N GENTAMICIN

THIOSTREPTON ..... .... ....

CHLORAMPHENICOL LINCOMYCIN S P A R S O N K I N

13110 Inhibition Mechanism of Aminoglycosides, Thiostrepton, and Viomycin

families (7) and are not considered here. The typical aminoglycosides consist of the neomycin family (including hygro- mycins), the kanamycin, and gentamicin families (33). All are glycosides of 2-deoxystreptamine. Streptomycin and dihydrostreptomycin are special in that they are derivatives of strep- tamine (34).

The structural relatedness of the aminoglycosides corre- sponds to a common resistance mechanism, since a base exchange or methylation around a narrow region of the secondary structure of the 16 S rRNA (positions 1405-1409 and 1491-1495 in the E. coli 16 S sequence) can confer resistance to kanamycin, gentamicin, paromomycin (a member of the neomycin family), and hygromycin (for review see Ref. 35). Streptomycin seems to be an exception, since a change from C912 to U can confer resistance against this drug (36, 37). However, dihydrostreptomycin could be cross-linked to a region near 900 and another one near 1400 (38). Since E. coli ribosomes have one binding site for this drug (39), one would expect these two regions to be spatially neighbored within the ribosome. In the 16 S rRNA model of Brimacombe et al. (40) this is indeed the case.

In excellent agreement with these data, the presence of neomycin, gentamicin, and kanamycin protects A1408 and G1494, and hygromycin protects G1494 against chemical modification (41). This indicates that all these antibiotics bind to the same region of the ribosome. Since the anticodon of a P site-bound tRNA could be cross-linked to C1400 (42), the aminoglycosides obviously bind at or near the decoding site, which has been mapped in the cleft between head and body of the 30 S subunit (43).

The surprisingly simple picture which arises from the common features of structure, genetic data, and biochemical evi- dence contrasts to the pleiotropic effects of the aminoglycosides that have been described (7): 1) irreversible uptake of the drug, 2) membrane damage, 3) misreading, and 4) ribosomal blockage. In any case the ribosomal target must play a key role, since a ribosomal mutation protects the bacterium from the lethal effects of the drugs, leaving either misreading or ribosomal blockade as the cause for the killing action.

The misreading could lead to a defective transcriptional and translational apparatus, thus amplifying the error rate and ending in an error catastrophe (44). However, experimental data do not favor this view. As mentioned by Fast et al. (45), ram mutations affecting ribosomal proteins S4 or S5 generate stable cultures but translate at high error rates. The authors show in the same paper that rate and accuracy of isolated ribosomes are hardly affected even if the cells were exposed to streptomycin at low concentrations.

Only the ribosomal blockage is left. Do the aminoglycosides inhibit one specific ribosomal function? They do not prevent the formation of the initiation complex and do not impede polysomes (6, 46). Obviously, they must block an elongating step shortly after initiation, but precisely which step was not a t all clear. Neomycin and hygromycin were shown to inhibit the translocation reaction (47, 48, and Fig. 2A), but streptomycin, kanamycin, and gentamicin hardly affect the translocation (66 to 89% residual activity, see Fig. 2A). Likewise, the effects on A site binding of the i-type or peptidyltransferase were weak (Fig. 2A and Table I). Notably, P site binding was not inhibited at all (Table I). Up to now, no specific inhibitor of this particular ribosomal function has been found.

However, as shown here all the aminoglycosides severely block one function of the elongating ribosome, namely the A- site occupation of the e-type (Fig. 2B). This seems to be the common point of interference and the primary inhibition function. Since polysomes are scarcely affected by aminogly-

cosides (retardation of the elongation rate probably due to an impairment of the e-type binding, see Fig. 4, and misreading), the drugs might bind preferentially to the 30 S subunit. Therefore, one can imagine the following steps leading to the ribosome blockage: the aminoglycoside binds to the 30 S subunit and allows the formation of the initiation complex (binding of mRNA, met-tRNA and the association of the 50 S subunit). The first elongation cycle occurs (filling of the E site), but then the A site occupation of the second elongation cycle is blocked. Members of the neomycin group might in addition already impair the first translocation reaction.

The interference points of the antibiotics described here, together with those of some other inhibitors of the ribosomal elongation cycle (for review see Ref. 6), are compiled in a functional scheme of the allosteric three-site model (Fig. 6).

Acknowledgments-We thank Drs. H.-J. Rheinberger and D. Schlessinger for many discussions, and Drs. H. G. Wittmann and R. Brimacombe for advice and support.

REFERENCES 1. Rheinberger, H.-J., and Nierhaus, K. H. (1986) J. Biol. Chem.

2. Liou, Y.-F., and Tanaka, N. (1976) Bbchem. Biophys. Res. Com- mun. 71,477-483

3. Modolell, J., and Vazquez, D. (1977) Eur. J. Biochem. 81, 491- 497

4. Vazquez, D. (1979) in Inhibitors of Protein Biosynthesis, pp. 52- 145, Springer-Verlag, Berlin

5. Cundliffe, E. (1981) in The Molecular Basis of Antibiotic Action (Gale, E. F., Cundliffe, E., Reynolds, P. E., Richmond, M. H., and Wering, M. J., eds) 2nd ed., pp. 402-547, John Wiley & Sons, London

6. Nierhaus, K. H., and Wittmann, H. G. (1980) Naturwissenschuf- ten 67,234-250

7. Davis, B. D. (1987) Microbwl. Rev. 51,341-350 8. Wurmbach, P., and Nierhaus, K. H. (1979) Methods Enzymol.

9. Rheinberger, H.-J., Schilling, S., and Nierhaus, K. H. (1983) Eur. J. Biochem. 134,421-428

10. Geigenmuller, U., Hausner, T.-P., and Nierhaus, K. H. (1986) Eur. J. Biochem. 161,715-721

11. Leberman, R., Antonsson, B., Giovanelli, R., Guariguata, R., Schumann, R., and Wittinghofer, A. (1980) Anal. Biochem.

12. Bartetzko, A., and Nierhaus, K. H. (1988) Methods Enzymol., in

13. Wurmbach, P., andNierhaus, K. H. (1983) Eur. J. Biochem. 1 3 0 ,

261,9133-9139

60,593-606

104,29-36

press

14.

15.

16.

17.

18.

19. 20.

21. 22.

23.

24.

25.

26.

27.

9-12 Bergemann, K., and Nierhaus, K. H. (1983) J. Bid. Chem. 2 6 8 ,

Rheinberger, H.-J., Sternbach, H., and Nierhaus, K. H. (1981)

Rheinberger, H.-J., Sternbach, H., and Nierhaus, K. H. (1986) J.

Lucas-Lenard, J., and Haenni, A.-L. (1968) Proc. Natl. Acad. Sci.

Geigenmuller, U., and Nierhaus, K. H. (1986) Eur. J. Biochem.

Traut, R. R., and Monro, R. E. (1964) J. Mol. Biol. 10.63-72 Cundliffe, E., and McQuillen, K. (1967) J. Mol. Biol. 30, 137-

Endo, Y., and Wool, I. G. (1982) J. Bwl. Chem. 257,9054-9060 Hausner, T.-P., Atmadja, J., and Nierhaus, K. H. (1987) Bio-

Rheinberger, H.-J., and Nierhaus, K. H. (1983) Proc. Natl. Acad.

Rheinberger, H.-J., and Nierhaus, K. H. (1986) FEES Lett. 2 0 4 ,

Rheinberger, H.-J., and Nierhaus, K. H. (1987) J. Biomolec.

Robertson, J. M., and Wintermeyer, W. (1987) J. Mol. Biol. 196 ,

Monro, R. E., and Vazquez, D. (1967) J. Mol. Biol. 2 8 , 161-165

15105-15113

Proc. Natl. Acad. Sci. U. S. A. 78, 5310-5314

Biol. Chem. 261,9140-9143

U. S. A. 59,554-560

161,723-726

146

chimie 69,911-923

Sci. U. S. A. 80,4213-4217

97-99

Structure Dynamics, 5,435-447

525-540

Inhibition Mechanism of Aminoglycosides, Thiostrepton, and Viomycin 13111

30. 31.

32. 33.

34.

35. 36.

37.

8833 Pestka, S. (1972) J. Biol. Chem. 247,4669-4678 Contreras, A., and Vbquez, D. (1977) Eur. J. Bwchem. 74,539-

Pestka, S. (1970) Bwchem. Bwphys. Res. Commun. 40,667-674 Cundliffe, E. (1986) in Structure, Function, and Genetics of Ri-

bosomes (Hardesty, B., and Kramer, G., eds) pp. 586-604, Springer-Verlag, New York

Nagabushan, T. L., Miller, G. H., and Weinstein, M. J. (1982) in The Amimglycosides (Whelton, A., and Neu, H. C., eds) pp. 3- 28, Marcel Dekker Inc., New York

547

Cundliffe, E. (1987) Biochimie 69,863-869 Montandon, P. E., Wagner, R., and Stutz, E. (1986) EMBO J. 6,

Etzold, Th., Fritz, C. C., Schell, J., and Schreier, P. H. (1987) 3705-3708

FEBS Lett. 219,343-346

40. Brimacombe, R., Atmadja, J., Stiege, W., and Schuler, D. (1988)

41. Moazed, D., and Noller, H. F. (1987) Nature 3 2 7 , 389-394 42. Ofengand, J., Ciesiolka, J., Denman, R., and Nurse, K. (1986) in

Structure, Function and Genetics of Ribosomes (Hardesty, B.,

43. Oakes, M. I., Clark, M. W., Henderson, E., and Lake, J. A. (1986) and Kramer, G., eds) pp. 473-494, Springer-Verlag, New York

Proc. Natl. Acad. Sci. U. S. A. 83,275-279 44. Orgel, L. E. (1963) Proc. Natl. Acad. Sci. U. S. A. 49 , 517-521 45. Fast, R., Eberhard, T. H., Ruusala, T., and Kurland, C. G. (1987)

46. Luzzatto, L., Apirion, D., and Schlessinger, D. (1969) J. Mol.

47. Cabahas, M. J., Vazquez, D., and Modolell, J. (1978) Eur. J.

48. Misumi, M., Nichimura, T., Komai, T., and Tanaka, N. (1978)

J. Mol. Biol. 199 , 115-136

Biochimie 69,131-136

BWl. 42,315-355

Biochem. 87,21-27

Biochem. Bwphys. Res. Commun. 43,196-199

the allosteric three-site model for the ribosomal elongation cycle

Documents